Professional Documents
Culture Documents
Richard A. Epstein*
149
which stipulates that state 2 is better than state 1 if and only if the winners in state 2 can
afford to compensate the losers fully and still remain better off themselves.
John Rawls's use of the "veil of ignorance" marks an effort to collapse the two tests. If
no one knows his place in nature then he will opt for that social order that will leave the
single h)potheticalchooser better off than any other. The point behind the method is that
it allows the analyst to escape the hold-out problem that otherwise exists when unanimous consent is needed to form a civil society. Yet by the same token it ignores the very
real differences in tastes and attitudes that surely exist in the world at large, and it provides no reliable guide as to what formal function identifies the ideal single person from
the welter of individuals in the world at large.
HeinOnline -- 56 Geo. Wash. L. Rev. 150 1987-1988
[VOL. 56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
1987]
particular people.8
Some rule of law is better than no rule of law. But the choice of the
best, even the best achievable, form of political organization demands more than faithful adherence to the rule of law can provide.
There are many, wholly different, legal systems that have all the
marks of the rule of law. It is very easy to tell which commands are
indeed those of the sovereign and which are bogus. The rules themselves may well be very general in both form and content. The principles used to organize Nazi Germany (an issue that rightly obsessed
Hayek) met in some formal sense the requirements of the rule of
law. The citizen could easily figure out that they were authored by
the sovereign and written with a high degree of generalization. A
rule that says that all Jews must wear a yellow star has the same
generality and clarity as one that says that all Jews are entitled to
exemption from military service on the Sabbath. But there the simi8. Id. at 74. Note too that Hayek also condemned the broad discretion conferred
upon English administrative agencies. See idat 62-63. This same theme was at issue in
the United States at the same time. Contrast with Hayek the following excerpts from a
Frankfurter Court opinion in National Broadcasting Co. v. United States, 319 U.S. 190
(1943):
The [Federal Communications] Act itself establishes that the [Federal
Communications] Commission's powers are not limited to the engineering
and technical aspects of regulation of radio communication. Yet we are
asked to regard the Commission as a kind of traffic officer, policing the wave
lengths to prevent stations from interfering with each other. But the Act
does not restrict the Commission merely to supervision of the traffic. It puts
upon the Commission the burden of determining the composition of that
traffic. The facilities of radio are not large enough to accommodate all who
wish to use them. Methods must be devised for choosing from among the
many who apply. And since Congress itself could not do this, it committed
the task to the Commission.
Id. at 215-16. Justice Felix Frankfurter then went on to hold that the "'public convenience, interest, or necessity" standard of the act furnished the Commission with sufficient guidance for its task. Id. at 216 (quoting 47 U.S.C. 303 (1940)). It is a striking
confirmation of Hayek's thesis that the licensing program of the Federal Communications Commission proved a total failure once it moved beyond its traffic-cop function.
The seminal criticism (in this small world) was by R.H. Coase, who noted that scarcity
was not a reason for government allocation, but for governments to define property
rights so that private (and public) parties could bid for the frequencies in a market system. See Coase, The FederalCommunications Commission, 2 J.L. & EcoN. 1,14 (1959).
152
[VOL. 56:149
HeinOnline -- 56 Geo. Wash. L. Rev. 152 1987-1988
Epstein
THE GEORGE WASHINGTON LAW REVIEW
1987]
law. Yet note what Hayek wrote: "To prohibit the use of certain
poisonous substances, to limit working hours or to require certain
sanitary arrangements, is fully compatible with the perservation of
competition."' 0 Yet the risks that so-called neutral health laws have
upon competition were a threat well understood at the time of Lochner v. New York," and they have been documented extensively in 2the
modem regulatory work dealing with agencies such as OSHA.1
There is therefore a clear need to go beyond the form of a law to
decide whether it is just or wise, and a normative theory of human
behavior and political irrtitutions is needed to explain why Dicey's
and Hayek's intuitions about markets and government power, for
example, are correct. There are no shortcuts in the process. Too
much weight therefore is placed upon the rule of law to filter out
good from bad legal rules. Hayek may have wished to contrast the
rule of law with substantive rules, but his distinction does not hold
true. All laws, general and specific, have an irreducible substantive
component. A rule of law that allows the sovereign, be it king or
legislature, to generate any clear command that it likes is not a very
attractive system, even if it is, as Hobbes argues, better than one in
which all forms of individual license are subject to no external constraint. The harder question therefore is how does one choose
among different types of legal orders, all of which satisfy the formal
requirements of the rule of law. The answer to that question again
drives us beyond the rule of law, to the foundations of political theory and modern constitutionalism.
II.
154
[VOL.
56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
subsequent democratic theorists who have rejected the case for his
undivided, absolutist sovereign. These theorists have sought to find
need for the sovereign to maintain a system of
ways to recognize ' the
"ordered liberty,"' 3 one that simultaneously skirts the risks of the
war of all against all and the alternative risks of vesting absolute
power in a monolithic, possibly despotic, Hobbesian sovereign.
The concern with Hobbes's own alternative rests with the very
sensible objection that although many absolute sovereigns may be
well-intentioned rulers, some will not. As a rough empirical generalization, the prosperity and tranquility that good rulers can afford
will not normally be sufficient to compensate for the reign of terror
that bad despots can impose. A bull market on the New York Stock
Exchange does not compensate for terror, imprisonment and death.
The issue then is what kinds of devices can be used to counteract
the authority of the sovereign once we move into civil society. In
our political literature two devices have been proposed, and these
explain the title of this Article. The first of these is some idea of
civic virtue, and the second is structural safeguards against the concentration of power. I shall talk about each briefly.
The idea of civic virtue is an elusive one, and I for one am not
quite sure I understand the complex constellation of ideas that lurks
beneath its surface. But whatever its precise contours, civic virtue
surely was an important concern in political theory both before and
during the framing of our constitution, and it has had a sudden return to prominence in modem academic circles. 1 4 The modem discussions of civic virtue have directed renewed attention to the work
ofJames Harrington, 15 who, writing in response to Hobbes, found
in civic virtue an antidote, imperfect perhaps, to unrestrained greed
and self-interest. Individuals must be raised in a civil society where
they are taught to understand that their own individual wills and
desires must be subordinated to those of the larger society. If many
individuals learn to behave in this way, then they can develop into a
sufficient critical mass, which will form an effective counterweight to
the arbitrary power of wayward government officials. Closely akin
to civic virtue therefore is the idea of extensive general "participa13. The phrase isJustice Benjamin Cardozo's in Palko v. Connecticut, 302 U.S. 319,
325 (1937), but the problem both precedes and survives his efforts to resolve it.
14. See Michelman, Foreword: Traces of Self-Government, 100 HARV. L. REV. 4 (1986);
Sunstein, Interest Groups in American Public Law, 38 STAN. L. REV. 29 (1985).
15. Harrington lived from 1611 to 1677, and his most famous work, The Commonwealth of Oeana, was written in 1656, five years after the appearance of Hobbes's Leviathan. Oceana was Harrington's mythical name for England. For a selection of
Harrington's writings, seeJ. HARRINGTON, THE POLITICAL WRITINGS OFJAMES HARRINGTON: REPRESENTATIVE SELECTIONS (C. Blitzer ed. 1955) [hereinafter HARRINGTON, W=inINGs]. For an exhaustive compilation of Harrington's work, see J. HARRINGTON, THE
POLITICAL WORKS oFJAMEs HARRINGTON (J. Pocock ed. 1977) [hereinafter HARRINGTON,
WORKS].
19871
155
there was a natural aristocracy, recognizable to all men, the members of which were endowed with more of the right stuff than
others. 17
[VOL. 56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
few may be the light of mankind, but the interest of the few is not the
'26
This preoccupation with self-interest as a threat to the body politic becomes especially clear from the decisive illustration that Harrington uses to derive the structural division between the senate
proposing and the people resolving. His basic analogy is to the
21. "An equal Agrarianis a perpetual law establishing and preserving the balance of
dominion by such a distribution that no one man or number of men within the compass
of the few or aristocracy can come to overpower the whole people by their possessions
in lands." HARRINGTON, WRrrINGs, supra note 15, at 71.
22. The restrictions included those that prevented a person who had land valued in
excess of 2,000 pounds both from leaving it to a single son where division among sons
was possible and from acquiring new lands, except by inheritance, valued in excess of
that same sum. See HARRINGTON, WORKS, supra note 15, at 62-63.
23. He unmercifully criticizes Hobbes, who had published Leviathan just five years
before, for his monarchial tendencies and his willingness to place the sovereign above
the law. See HARRINGTON, WRITINGS, supra note 15, at" 60-61, 67, 69.
24. Id. at 71-72. This sentence links together both Harrington's broad social con-
1987]
given her knowledge that the other girl will choose the larger part if
the cake is not divided in equal halves. The self-interest of each side
is now made to operate under a set of constraints that dictates
movement toward the socially desirable outcome.2 9 There is not
only the rule of law, but there is also the right rule of law.
Yet note that there are substantial difficulties that lurk in the use
of this game for understanding divided sovereignty. In the first instance the game need not always yield the right result. Assume, for
example, that some error is possible both in the cutting of the cake
and the judging of the size of the pieces. The first alone means that
the two pieces are not of equal size, so that the imperfect ability to
execute a mechanical task will in part frustrate the ideal of equal
division, self-interest notwithstanding. In addition, given the presence of error costs, it now makes a difference whether a given person goes first or second. Here we can assume, empirically, that it is
harder to cut than to judge, just as it is harder to play the piano than
to compare the talents of two pianists. It follows therefore that each
of the two participants in the game has a clear incentive to do the
judging and not the cutting. It is highly likely both that the two
pieces will not be of equal size and that the second girl will be able
to spot the difference and choose accordingly. The easy benevolence of Harrington's dialogue will not survive the rigors of the
world. Indeed if the game is a one-shot affair, some inequality will
have to be accepted, even if lot is used to determine which girl goes
27. Id. at 58.
28. Id.
29. J.G.A. Pocock says this game "ensures that neither has an interest in behaving
unjustly, with the consequence that each girl helps the other maintain the supremacy of
reason over passion. Each not only gets her fair share of the cake, but receives the
greater good of having acted virtuously." HARRINGTON, WORKS, supra note 15, at 65.
Pocock's formulation places too much emphasis upon virtue and reason. The point of
the game is to make sure that self-interested players get the right result even when they
act out of passion and not virtue. The virtue belongs to the person who designed the
game, not to the girls who play it.
HeinOnline -- 56 Geo. Wash. L. Rev. 158 1987-1988
[VOL. 56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
first. Ironically, however, if the game is repeated, then the girls can
take turns going first, so that on analogy by Harrington's own principle of rotation they can approach equality in the long term. The
lesson is important, for with imperfect information, the perfect incentive-compatible rule (that is, the first player in the exercise of his
own self-interest does that action that advances the social interest)3 0
does not yield the ideal result. But lest we despair, this rule is still
clearly superior to alternatives that might be put into its place.
The second point is that the simple example does not easily generalize to games that involve more than two people. Thus suppose
one wants to use the rule to divide a cake into three equal slices.
The only way that this can be achieved is to have one player make all
the cuts, and then to allow the other players, in an order chosen at
random, to select their pieces, leaving to the cutter the last piece. If
there are no side agreements among the players, the first player
knows that if any one piece is smaller than the rest, she will get it, so
the incentive is to keep all pieces the same size. This procedure is
subject to all the caveats about error costs raised above, and has in
addition the odd feature that only one person can do the cutting if
the game is to work at all.
The moment two or more players are allowed to cut, the possibilities of strategic misbehavior are clearly introduced. Take a game
with three players and three slices of cake, in which the first to cut is
the last to pick, the second to cut is the second to pick, and the third
player does not cut, but picks first. If the first player cuts the perfect
third, then the second player (who will pick second) is undisciplined
in the way in which she wields the knife. She can cut unequal slices
if she so chooses, knowing that she can always take the one-third
already cut, leaving to the last player (who will pick first) the largest
slice of the cake. The expanded game therefore severs authority
from the consequences of authority and always places the first
player at the mercy of the second - the very result Harrington
feared.
This difficulty cannot be forestalled by having the first player cut a
slice either greater or less than the perfect third. If the slice is cut
larger, then the third player will pick it unless the second division is
so unequal that it yields a still larger piece. If the first player cuts a
slice smaller than a third, then the even division of the uncut portion
will yield both the second and third players larger shares than the
first. These possibilities of strategic behavior multiply when the
game goes from 3 to n players. The only rule that tends toward
equal slices is one that directs the first player to cut all of the slices.
30. For a further elaboration of the role of this principle in the general law of tak-
1987]
TAK-
33. The same pattern can be found in modern administrative law, where dominance
of the New Deal view has meant that there are no vested property rights that are immune
from comprehensive social regulation. That having been decided, the emphasis then
shifts to the administrative procedures that might be introduced to constrain the behavior of the legislators and administrators who operate within the capacious confines of
the majority will. But in general these efforts will fail precisely because they are not
animated by any substantive vision of what the just society looks like. For the most
comprehensive effort to rationalize the difficulties in modern administrative law, see
Stewart, The Reformation of American Administrative Law, 88 HARV. L. REv. 1667 (1975).
160
[VOL. 56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
worse for the other in case she had not divided at all but kept the
whole cake to herself, in regard that, being to choose too, she divided accordingly. Wherefore if the senate have any further
34
power than to divide, the commonwealth can never be equal.
The contrast between this structure and the two-party cake game
is stark and complete. The cake game was designed to be immune
from all possibility of strategic behavior and achieves just that result
if the cake can be cut precisely in two. The division between the
senate and the people opens up vast horizons on which indeterminate political struggles - bilateral monopoly problems again - can
take place, even on Harrington's own relatively simple model in
which twenty representatives divide themselves (it is never dear
how) into six senators and fourteen representatives for the people.
Initially there can be disagreement among the senators over what to
propose or among the people over what to accept. Even in the unlikely event that both the senate and the people are unanimous in
their separate views, there could be serious struggles between them,
as each has a blocking position against the other. In the cake game
there is a clear advantage to being the chooser and not the cutter.
Yet it is unclear whether the people have greater power than the
senate, and probably it is likely that the reverse is true, given that
the people, even if unanimous, will not reject a senate proposal if
their own costs in getting a better proposal adopted exceed any advantage that they hope to achieve from jockeying. 35 Harrington's
probable implication from the cake game - that an asymmetrical
division of power between two branches of a legislature can work as
smoothly, or even nearly as smoothly, as the cake game - is wholly
incorrect. Yet here, as everywhere in political life, the best cannot
be the enemy of the good. Any criticism of Harrington's analogy
between cakes and governments does not show that the division of
sovereign power is unwise or unattainable. Indeed we know from
34. HARRINOTON, WRITINGS, supra note I5, at 59-60.
35. Thus take a game in which there is an array of different canned goods. One
player can break them into two piles and the second player picks the pile of his choice.
Is it better to divide or to pick? Here, if the first player knows the subjective preferences
of the second it is better to divide. The winning strategy exploits the difference between
market and subjective value. Thus the first player creates two piles. One pile may have a
market value of 45, a subjective value of 40 to that first player, and a subjective value of
60 to the second player. The second pile may have a market value of 55, a subjective
value of 70 to the first player, and a subjective value of 55 to the second player. If we
rule out the possibility of trade after the division, then the second player will take that
package with the higher subjective value to her (60 to 55), that is, the first pile, even
though the first player then is left better off than the second (70 to 60). If there is the
possibility of trade after the division, then the second player may take the other pile and
then hold out to secure the goods she desires most. Note also that in the limiting case
where the subjective values of the two players are identical, the first player cannot win.
Yet in practice the difference between the objective and subjective values is likely to be
large. If two patrons ofa supermarket each spend fifty dollars on groceries, neither will
be pleased if she takes home the groceries of the other by mistake.
1987]
III.
ConstitutionalRepublicanism
[VOL. 56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
1987]
talks about the two strains from Harrington, civic virtue and structural restraints. Sensibly enough he is not prepared to rely upon
one to the exclusion of the other. On civic virtue he first expresses
the hope that a system of distance and indirect election will increase
the likelihood of selecting legislators who will be able to place the
common good over the special interests of factions. Thus the rules
for the election of the senators, with their long terms and appointment by state legislators, certainly are designed to keep the members of one house of Congress insulated from the popular pressures
of the day. Yet the fear of self-interest that moved Harrington (and
Locke) troubled Madison as well, for he knew full well that
"[e]nlightened statesmen will not always be at the helm."140 Civic
virtue is nice when you can get it, and in the long run a nation that
places a succession of fools and knaves in high places will not be
able to survive unscathed regardless of its constitutional structure.
But politics, as a game of playing the odds, is devoid of certainties.
The true test of government is how well, and how long, its institutions will be able to withstand the misdeeds of bad leaders in difficult times. Here Madison's hunch is the assertion that his
"extended republic" will be at a sufficient distance from the people
that it will be able to avoid the vices of excessive familiarity while not
being wholly aloof and uninformed about the needs and desires of
the public that it is obliged to protect.
In one sense Madison is very much like Hobbes. His basic account of the problem of factions is far more persuasive than his proposed cure, just as Hobbes's account of the state of nature is more
powerful than his monolithic solution. Madison may be right that,
over the full array of possible cases, distance is better than no distance. But this superiority translates itself only into the unhappy
conclusion that it takes the extended republic little more time to
decline into the popular despotism that he so fears. The gist of
Madison's difficulty is that the formal constraints of his extended
republic - complex representative government, working at a distance - are simply unequal to the challenge of the incessant pounding of self-interest. In a sense he probably knew the point as well as
anyone else, for so much of the rest of The FederalistPapersis directed
to identifying and justifying other limitations that the Constitution
imposes upon the federal government.
I too am unable to propose a set of foolproof institutional arrangements, so I shall content myself with pointing out the real
shortcomings in Madison's conception, treating for the moment The
FederalistNo. 10 as a self-contained essay, and not as a constituent
part of a larger work. In making these remarks, I should not be
understood as being critical of any effort to decentralize power by
placing it in the hands of many different sovereigns as federalism
requires.4 1 Even though important mistakes in allocation can and
40. Id. at 60.
41. I discuss some of these points in far greater depth in Epstein, The Proper Scope of
the Commerce Clause, 73 VA. L. REv 1387 (1987).
164
[VOL.
56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
165
[VOL. 56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
drawn into the factional fray once the number of defections increases. The best any system of politics can generate is a world in
which some legislators will seek factional gains regardless of what
others do, where others seek factional gains only if some people
have done so first, where still others will do so only if many others
have done so before them, and so forth. Yet given that all too plausible distribution, it should take only a little time before each legislator has to act out of narrow self-interest to keep up with his rivals.
Because each legislator has to determine for himself whether others
have misbehaved, it is quite unlikely that all legislators will hew to
the virtuous line.
Legislatures are quite unlike ordinary product markets. There the
shifty player will be excluded over time as other traders will look
elsewhere for honest partners. But in politics no one can be ignored because simple two-party contracts do not pass laws. Instead
laws are common pool assets whose stability is as fragile as that of
the common pools in oil and gas and water rights. 44 The pressures
on virtuous legislators in a system of majority rule over time should
prove very strong indeed. Complex systems of voting at a distance,
then, suffer the same fate as the rule of law. They clearly are indispensable in any system of government, but they are hardly sufficient
to control the array of pressures that modem political markets can
generate. The pessimism of the dogged Hobbesian is not displaced
by Madison's institutional arrangements.
IV. Controlling the Factions
The question then arises; what other elements are needed to improve the situation? Here it is puzzling that Madison in his effort to
defend the extended republic did not mention in The Federalist No.
10 a number of features found in our Constitution that help to
dampen the influence of factions. Thus in the original Constitution
there was the indirect election of senators on staggered six-year
terms, which could work to weaken the connection between popular
sentiment and senatorial behavior. The sentiments of a given moment could never allow the electorate at a single time to constitute a
majority of one house of Congress. Yet by the same token the constant reelection of the House of Representatives means that no legislation can pass unless it has also the support of legislators just
recently elected. The need to go through two filters, both the
Senate and the House, can only slow down legislation, a sensible
rule when the risk of too much legislation is thought greater than
44. For a discussion of these issues, see Epstein, Why Restrain Alienation?, 85 COLUM.
1987]
there is a political pie to be divided, the good senator sees that his
constituents get at least their fair share, and perhaps a little more.
Senators (even before direct elections) and congressmen therefore
are not chosen for their skill in putting brakes on the total level of
federal spending and regulation. But citizens who want a favorable
division of the political pie may also want to have a smaller pie to
divide in the first place, and therefore they could also vote for a
president who is more conservative (that is, who believes in a
smaller federal government). If senators and representatives are
chosen because they play the factional game well, presidents could
be chosen because they reduce the stakes for which the game is
played. The same voters therefore, often will have different preference functions for different offices, which makes it more difficult for
factions to operate.
The system of enumerated powers, also not discussed by Madison
in The Federalist No. 10, worked in the same direction until it was
overrun by the extravagant interpretations given to the commerce
clause by the decisions of the post-1987 New Deal Court.46 But
faithfully enforced, it is an important part of the total political equation. The power of factions to operate depends upon the gains that
faction members can hope to achieve through political action.
Where structural devices limit the returns to factional behavior, we
should expect to see fewer factions form. Where certain objects are
kept from the political process at the federal level, legislators have
fewer degrees of freedom over which they can deal. Just as the complex voting rules raise the costs of forming a coalition, so too the
respect for enumerated powers tends to limit the gains that any successful coalition can achieve.
These safeguards against factions are hardly perfect. Indeed as
45. Again the intuition here is quite strong. If there are certain public-interest-type
statutes that benefit everyone, for example, statutes of limitation, they will pass under
very stringent voting rules. Factional legislation will not command such broad support,
so that the additional constraints are more likely to trap it than public interest statutes.
46. See, e.g., Wickard v. Filburn, 317 U.S. 111 (1942) (stating that farm products
produced for home consumption are subject to federal regulation where they affect the
quantity of goods in interstate commerce); United States v. Darby, 312 U.S. 100 (1941)
(upholding regulation by Congress of the production of goods that manufacturers intend or expect to move in interstate commerce, regardless of whether those goods are
actually shipped in interstate commerce); NLRB v. Jones & Laughlin Steel Corp., 301
U.S. I (1937) (holding that the National Labor Relations Board was justified in controlling industrial employee/employer relationships in order to prevent a burden on or obstruction of interstate commerce).
168
[VOL. 56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
1987]
169
knowledge would take property only where they believed that the
gain to the rest of society would exceed the losses to the individual
property owner. That person in turn could protect himself against
unanticipated loss, if he so desired, by the purchase of insurance in
the private market. The state would be spared the onerous task of
valuation, and this cost saving would redound to all.
[VOL. 56:149
Epstein
THE GEORGE WASHINGTON LAW REVIEW
1987]